Computer Engineering and Applications ›› 2016, Vol. 52 ›› Issue (6): 227-234.

Previous Articles     Next Articles

Flintstone inspection and recognition based on illumination invariant perceptual feature

CHEN Wenming1, LIANG Dongtai1,2   

  1. 1.Faculty of Mechanical Engineering and Mechanics, Ningbo University, Ningbo, Zhejiang 315211, China
    2.Zhejiang Province Key Laboratory of Part Rolling Technology, Ningbo University, Ningbo, Zhejiang 315211, China
  • Online:2016-03-15 Published:2016-03-17

基于光照不变感知特征的打火石检测识别

陈文明1,梁冬泰1,2   

  1. 1.宁波大学 机械工程与力学学院,浙江 宁波 315211
    2.浙江省零件轧制成形技术研究重点实验室,浙江 宁波 315211

Abstract: In order to recognize the direction of flintstones and improve the efficiency of installation, a color machine vision system is proposed to inspect the two ends of flintstones according to the colors of their two end faces, where a chessboard like device is used to hold the cylinder-shaped flintstones in rows and columns. For the accurate detection of flintstones’ magenta end faces, a perceptual color feature recognition algorithm is proposed based on color opponent process theory. And illumination invariant features are obtained by feature transformation to reduce the influence of uneven illumination. For the detection of the other silver end faces, Otsu threshold segmentation method is used and obvious bright color feature can be extracted effectively. In order to avoid inspection fault, the statistical relationship of color channels is further used to limit the color feature extraction of silver end faces. The experimental results show that the illumination invariant perceptual feature based on opponent color theory and the threshold segmentation method can effectively recognize the color features of flintstones’ end faces. And the accuracy rate for the magenta end faces reaches up to 99.87% and the accuracy rate for the silver end faces reaches up to 99.62%, which meet the performance requirements for the visual inspection.

Key words: color feature extraction, illumination invariant feature, opponent process theory, color recognition

摘要: 为了在生产过程中确定打火石的方向,提高打火石安装定位的效率,根据打火石圆柱状外形及两端面的颜色特征,使用特定的棋盘形工装定位装置对打火石进行定位,并利用彩色机器视觉技术对打火石的端面颜色进行检测识别。为了准确识别出打火石的品红色端面,在拮抗色感知特征基础上,针对现场环境中光线不均等问题,进行了光照不变的拮抗色感知特征变换,降低了光照不均对检测结果的影响。对于打火石另一端面银白色特征,根据其高亮的特性,采用Otsu自适应阈值法进行分割,并在颜色通道的统计基础上,对银白色特征提取过程进一步进行约束。实验结果表明,采用光照不变的拮抗色感知特征提取算法以及阈值分割相结合的方法,能够有效对打火石的端面颜色进行识别,在不出现误检的情况下,品红色端面的识别准确率为99.87%,银白色端面的识别准确率为99.62%,满足了打火石视觉检测识别系统的性能要求。

关键词: 彩色特征提取, 光照不变特征, 拮抗色理论, 颜色识别